Ethernet networks can be realized using meshed or ringed network architectures. The use of wireless technology in the core and access links of such networks is gaining momentum because of the ease of installation of such links and the associated lower installation cost. One driving factor of post-installation cost, however, is the bandwidth offered by wireless links. While per link bandwidth offered by wireless links is controlled by varying factors such as radio spectrum in use, coding and modulation techniques used, etc., some of these factors are regulated by governing bodies. Customers using these links are therefore often faced with paying for the full bandwidth offered by the links even though their actual bandwidth requirements based on amounts of communication traffic could be much lower than the full bandwidth.
One way to mitigate this problem is to initially purchase a certain amount of bandwidth, to which communication traffic will be rate limited. This allows a customer to pay for an amount of bandwidth that is estimated to be sufficient for current requirements, instead of the full bandwidth offered by wireless links. Additional bandwidth can then be purchased as the need arises.
However, this type of mechanism may pose problems, especially in larger networks. For example, each wireless link might need to be individually monitored by the customer to check its bandwidth utilization. For this purpose, the customer would need some sort of mechanism to continuously monitor the bandwidth requirement of each link, which of course would be very tedious in a large network. Also, since bandwidth is limited to the amount that the customer originally purchased, rate-limited links might not be able to support bursty traffic which exceeds the pre-purchased bandwidth. This results in links discarding excess traffic.
A pre-purchase and rate-limiting mechanism may also create difficulties when additional bandwidth is needed, such as where links must be individually upgraded if higher bandwidth is required.